Multi-agent communication system with automatic agent ID detection.
Setup
1. Team Information
cp examples/team.json.template /root/.openclaw/team-bus/team.json
Team members:
| Code name | AgentID | Responsibilities |
|---|---|---|
| Prism | lead | Coordination, aggregation, Telegram communication |
| Scope | product | Requirements, PRD, User Stories |
| Pixel | coder | Code implementation, bug fixes, testing |
| Lens | architect | Architecture design, interface definition, code review |
| Shutter | ops | Deployment, CI, integration testing, security monitoring |
2. Install Skill
cd <agent-workspace>/skills
git clone https://github.com/louis-cai/openclaw-team-bus-skills.git openclaw-team-bus
3. Configure HEARTBEAT
cp examples/HEARTBEAT.template <workspace>/HEARTBEAT.md
Usage
python3 bus.py <command> [args]
Commands:
send <agent> <title> <desc> <chat> --from <agent> # 发送任务（必传）
poll # 扫描收件箱（自动获取agent ID）
reply <agent> <task-id> <msg> [--accountId <id>] # 回复（accountId 必传）
broadcast <msg> [--chatId <id>] [--accountId <id>] # 广播（分发到每个 agent 的 inbox）
list-agents # 列出 agent
team # 显示团队信息（我是谁）
complete <task-id> [result] # 完成任务
fail <task-id> <error> # 标记失败
Agent ID
Agent ID is automatically obtained from environment variables:
TEAM_BUS_AGENT(manual configuration)CLAW_AGENT_ID(automatically provided by OpenClaw)
No manual input is required; commands such as poll/team automatically identify their own identity.